Animal Behaviour 101 (2015) 141e154

Contents lists available at ScienceDirect

Animal Behaviour

journal homepage: www.elsevier.com/locate/anbehav

Listening carefully: increased perceptual acuity for species discrimination in multispecies signalling assemblages

* Anna Bastian , David S. Jacobs

Department of Biological Sciences, University of Cape Town, Cape Town, South Africa article info Communication is a fundamental component of evolutionary change because of its role in mate choice Article history: and sexual selection. Acoustic signals are a vital element of animal communication and sympatric species Received 1 August 2014 may use private frequency bands to facilitate intraspecific communication and identification of con- Initial acceptance 23 September 2014 specifics (acoustic communication hypothesis, ACH). If so, animals should show increasing rates of Final acceptance 11 November 2014 misclassification with increasing overlap in frequency between their own calls and those used by Published online sympatric heterospecifics. We tested this on the echolocation of the horseshoe bat, Rhinolophus capensis, MS. number: 14-00625R using a classical habituationedishabituation experiment in which we exposed R. capensis from two phonetic populations to echolocation calls of sympatric and allopatric horseshoe bat species (Rhinolophus Keywords: clivosus and Rhinolophus damarensis) and different phonetic populations of R. capensis. As predicted by acoustic assemblages the ACH, R. capensis from both test populations were able to discriminate between their own calls and acoustic communication calls of the respective sympatric horseshoe bat species. However, only bats from one test population acoustic communication hypothesis fi bats were able to discriminate between calls of allopatric heterospeci cs and their own population when both echolocation were using the same frequency. The local acoustic signalling assemblages (ensemble of signals from functional extension sympatric conspecifics and heterospecifics) of the two populations differed in complexity as a result of habituationedishabituation contact with other phonetic populations and sympatric heterospecifics. We therefore propose that a receivers' perception acuity hierarchy of discrimination ability has evolved within the same species. Frequency alone may be suffi- Rhinolophus cient to assess species membership in relatively simple acoustic assemblages but the ability to use species discrimination additional acoustic cues may have evolved in more complex acoustic assemblages to circumvent mis- identifications as a result of the use of overlapping signals. When the acoustic signal design is under strong constraints as a result of dual functions and the available acoustic space is limited because of co- occurring species, species discrimination is mediated through improved sensory acuity in the receiver. © 2015 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.

Communication plays a crucial role in almost all aspects of an originate and are modified over evolutionary time is therefore animal's life (e.g. Maynard Smith & Harper, 2003) and is especially crucial to our understanding of the processes that generate biodi- important for species discrimination (Bradbury & Vehrencamp, versity (Mendelson & Shaw, 2012). It is likely that communication 2011; Ryan & Rand, 1993). It transmits information within a spe- systems evolved from systems used for other purposes (Monteiro & cies as well as across species and may have evolved as a product of Podlaha, 2009; Tinbergen, 1952), such as the function of feathers species coexistence (Li et al., 2013). Discriminating species is first used for insulation being extended so that they also function as important in interactions with heterospecifics allowing identifica- visual signals, for example in courtship displays (Cowen, 2005). tion of competitors, predators and prey, whereas the recognition of Particularly, knowledge of processes involved in the evolution of conspecifics is a prerequisite for any species-specific interactions, dual functions for a single trait can provide insight into how especially for mate choice (Jones, 1997; Sandoval, Mendez, & phenotypic diversity in both form and function is generated from Mennill, 2013; Slabbekoorn & Smith, 2002; Wilkins, Seddon, & existing variation. Safran, 2013). Understanding how communication signals Echolocation may provide us with an opportunity to investigate such functional extension of a trait. Echolocation is primarily used for orientation and food acquisition in echolocating bats, birds and & & * whales (Brinkløv, Fenton, Ratcliffe, 2013; Schnitzler, Moss, Correspondence: A. Bastian, UCT, Upper Campus, Private Bag X3, Rondebosch, & 7701, Cape Town, South Africa. Denzinger, 2003; Thomas, Moss, Vater, 2004) but there is E-mail address: [email protected] (A. Bastian). increasing evidence that it also functions as a means of http://dx.doi.org/10.1016/j.anbehav.2014.12.010 0003-3472/© 2015 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. 142 A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 communication (Gregg, Dudzinski, & Smith, 2007; Jones & Siemers, 1989; Jacobs et al., 2007; Kingston, Jones, Zubaid, & Kunz, 2000; 2010). In the context of species discrimination, communication Kingston & Rossiter, 2004; Russo et al., 2007). This acoustic diver- cues have to be unambiguous and represent a reliable badge for the gence among signallers has also been found in morphologically species. Such species-specific cues are present in the vocalizations cryptic species living in sympatry (Guillen, Juste, & Ibanez,~ 2000; of many animal groups including insects, anurans, birds and Jones & Siemers, 2010; Jones & Van Parijs, 1993; Kingston et al., mammals (primates: Seyfarth, Cheney, & Marler, 1980; anurans: 2001; Thabah et al., 2006). However, a test of the ACH would also Duellman & Pyles, 1983; birds: Catchpole & Slater, 2008; insects: have to incorporate an investigation of the perception and Pennetier, Warren, Dabire, Russell, & Gibson, 2010). Vocalizations discrimination ability of the receiver. are often a crucial signal in mate choice (Anderson, Ambrose, In this study we used the horseshoe bat, Rhinolophus capensis,to Bearder, Dixon, & Pullen, 2000; Braune, Schmidt, & Zimmermann, investigate the role of echolocation in communication in the 2008; Charlton, Huang, & Swaisgood, 2009; Vannoni & context of the ACH. We chose a classical habituationedishabitua- Mcelligott, 2007) as they can provide information about the tion experiment (Eimas, Siqueland, Jusczyk, & Vigorito, 1971)in sender which is used by the receiver to evaluate the mate's inten- which we exposed R. capensis to recorded calls of two sympatric tion, compatibility and quality. The voice of mammals, for example, horseshoe bat species (Rhinolophus clivosus and Rhinolophus dam- is often an honest cue which allows an individual to assess the body arensis) and different phonetic populations of R. capensis. In these size or mass of the sender (Fitch, 2006; Liebermann & Blumstein, assemblages we have populations of the same species using 1991). Among echolocating mammals bats are ideal candidates different echolocation frequencies as well as different hetero- for studies on echolocation in the context of communication specifics using overlapping frequencies. This natural system pro- because most species form groups with complex social structures vides an excellent opportunity to test whether R. capensis (Kulzer, 2005) in which many interactions are managed by acoustic discriminates between different species and populations on the signals (Altringham & Fenton, 2003; Fenton, 1985). The acoustic basis of their echolocation calls. If acoustic divergence in the structure of their echolocation calls has a complex frequencyetime echolocation frequencies of R. capensis is a result of selection contour and there are many different types of calls (Maltby, Jones, & favouring the use of private frequency bands as proposed by the Jones, 2010) providing sufficient variation to encode multiple cues. ACH, R. capensis should show increasing rates of misclassification Furthermore, echolocation calls contain diagnostic information with increasing overlap between its own calls and those used by about the sender which can be useful for others and, as a frequently sympatric heterospecifics. This concomitantly means that in- available signal, echolocation transmits information free of addi- dividuals of R. capensis from the different phonetic populations tional costs to a receiver (Dechmann, Wikelski, Noordwijk, Voigt, & should have difficulty recognizing each other as belonging to the Voigt-Heucke, 2013). In echolocating bats, the relationship be- same species if they use calls of dissimilar frequency. In addition tween echolocation call frequency and body size is well established this system allows us to test whether peak frequency is the only (Jacobs, Barclay, & Walker, 2007; Jones, 1999), and echolocation parameter used by bats to discriminate between species. calls often carry species-specific signatures, individual signatures, population-specific signatures and sex-specific signatures (Jones & METHODS Siemers, 2010). Several recent playback studies have provided evidence that Study Animal conspecific bats are able to extract information encoded in the echolocation calls of other bats such as species membership, fa- Rhinolophus capensis (Cape horseshoe bat) has a wide distribu- miliarity and sex (Dorado Correa, Goerlitz, & Siemers, 2013; tion along the coastal belt of South Africa's Cape (Monadjem, Taylor, Knornschild,€ Jung, Nagy, Metz, & Kalko, 2012; Schuchmann, Cotterill, & Schoeman, 2010). This species emits resting frequency Puechmaille, & Siemers, 2012; Voigt-Heucke, Taborsky, & Dech- echolocation calls (RF, calls emitted by rhinolophid bats when mann, 2010) and have recently revealed a role in mate choice stationary and hunting from a perch; Neuweiler et al., 1987; (Puechmaille et al., 2014). However, because echolocation has Schnitzler, 1968) that vary by more than 10 kHz across its distri- evolved primarily for orientation and food acquisition (Schnitzler bution range (Fig. 1; Odendaal, Jacobs, & Bishop, 2014). The lowest et al., 2003) species assemblages that are composed of ecologi- resting peak frequency, 75 kHz, is found in the northwestern part of cally similar bat species, and which therefore have similar echolo- its distribution, and the highest, 86 kHz, in the southeast. These cation call structures (Denzinger & Schnitzler, 2013), should phonetic populations co-occur with various other horseshoe bat partition the acoustic characteristics of their echolocation calls so species, namely R. damarensis (Jacobs et al., 2013) in the northwest that the calls retain their species specificity (Duellman & Pyles, and R. clivosus in the southern and eastern part (Jacobs et al., 2007). 1983; Heller & von Helversen, 1989). The concept of acoustic When both juveniles and adults are considered, R. clivosus calls at divergence of signals for species discrimination in multispecies 87e92 kHz (Jacobs, n.d.) and R. damarensis at 79e87 kHz (Jacobs assemblages to avoid misidentification as a result of the use of et al., 2013). In both cases R. capensis populations echolocate on confusingly similar calls (Amezquita, Flechas, Lima, Gasser, & Hodl,€ average 2e9 kHz lower than the respective sympatric hetero- 2011; Tobias, Planque, Cram, & Seddon, 2014) is well established in specific. This natural system provides an excellent opportunity to animal communication (Grant & Grant, 2010; West-Eberhard, test the ACH. 1983). This idea was first advanced by Duellman and Pyles (1983) for anurans and Heller and von Helversen (1989) for bats and Study Sites later named the acoustic communication hypothesis (ACH, Jacobs et al., 2007) which is similar to the spectral partitioning hypothe- The experiments were done at two sites in South Africa: De sis coined by Amezquita et al. (2011). Both the ACH and the spectral Hoop Nature Reserve on the southern coast of South Africa (March partitioning hypothesis propose that sympatric animal species each 2012 and October 2012) which represents a geographically central uses a ‘private frequency channel’ to facilitate intraspecific population with RFs at 85 kHz and a second more remote popula- communication and identification of conspecifics (Heller & von tion at the Orange River near Lekkersing in the extreme north- Helversen, 1989). This is supported by the divergence in the western corner of South Africa on the border with Namibia echolocation frequency of some bat species possibly as a conse- (November 2012). The latter population uses considerably lower quence of the presence of other species (Heller & von Helversen, RFs of 75 kHz. A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 143

10° E 20° E 30° E 40° E molitor larvae) to increase motivation. Immediately after the ex- periments we fed them with 30e50 mealworms and again gave them water. To ensure high nutrition of the diet the mealworms were fed with fresh vegetables and fruit, dry dog food and mineral powder for 4 weeks prior to field trips. 20° S Experimental Approach

We did a habituationedishabituation playback experiment to Rda 85 Rcl 92 test whether R. capensis individuals from two different localities in South Africa, Lekkersing (LS) in the northwest and De Hoop (DH) in RcaLS 75 * the south, perceive differences in the acoustic structure of echo- 30° S 81 N location calls from different phonetic populations of their own 81 82 87 species and other sympatric and allopatric horseshoe bat species 83 85 RcaTF 86 0 500 (i.e. R. damarensis and R. clivosus). This experiment is well suited to 85 RcaHH 84 *RcaDH 85 test whether two or more stimuli classes are perceived by the study km subjects as being different and thus discriminated (Bouton, 2007; & Figure 1. Map of southern Africa indicating the distributions of three horseshoe bat Simmons, Popper, Fay, Gerhardt, 2003). One stimulus class was species, Rhinolophus damarensis in the west (Rda, dashed line), R. clivosus in the east played back continuously until the study subject no longer reacted (Rcl, dotted line) and R. capensis in the south (Rca, solid line). Circles indicate sample to it, i.e. it was habituated to this stimulus. Habituation is defined as sites for R. capensis, whereas the two asterisks refer to the two study sites. Numbers a gradual decline in response following repeated exposure ¼ ¼ ¼ indicate peak resting frequencies. LS Lekkersing, HH Heidehof, DH De Hoop, & TF ¼ Table Farm. The distribution of R. capensis is adapted from Odendaal et al. (2014) (Bukatko Daehler, 2003; Fantz, 1964). This is needed to have a including the additional resting peak frequencies across the range. The distribution of common zero baseline of activity across individuals against which R. damarensis is adapted from Jacobs et al. (2013) and the distribution from R. clivosus to evaluate the discrimination response. Once the bat was habitu- from Monadjem et al. (2010). ated a test stimulus was presented. If the bat became active again after switching to the test stimulus, it indicated a dishabituation Ethical Note (Bukatko & Daehler, 2003). Dishabituation is defined as a rapid recovery from habituation (Rankin et al., 2009) based on the ability The treatment of animals in this study met the requirements of to discriminate between the two stimuli classes. ethical guidelines of the Federation for Laboratory Animal Science Our set of stimuli consisted of six acoustic stimuli classes each Associations (FELASA), the American Society of Mammalogists and representing one of the four phonetic populations of R. capensis as the South African National Standard (SANS, 2008; Guillen, 2012; well as one of the other horseshoe bat species, R. clivosus and Sikes & Gannon, 2011). All experiments were done with permis- R. damarensis, each co-occurring with one of our R. capensis test sion of local authorities (2399/2102; 0035-AAA007-00081) and the populations. More precisely, our six playbacks contained calls from Animal Ethics Committee of the University of Cape Town (2012/ (1) R. capensis from De Hoop which echolocates at 85 kHz (abbre- V33/DJ) and conducted by a trained person (A.B., FELASA-B viated as RcaDH85) in the south, (2) R. capensis from Heidehof certificate). (RcaHH84) also in the south, (3) R. capensis from Table Farm (RcaTF86) in the southeast, and (4) R. capensis from Lekkersing Capture, Release, Housing and Animal Care (RcaLS75) in the northwest, as well as from (5) R. clivosus which echolocates at 92 kHz (Rcl92) and co-occurs with RcaDH85 at De Bats were caught with mist nets (Ecotone monofilament, Hoop, and (6) R. damarensis echolocating at 85 kHz (Rda85) and co- 20 20 mm mesh size, Avinet, Dryden. NY, U.S.A.) at the entrance occurring with RcaLS75 at Lekkersing). Figure 1 shows a map of to two caves (Guano Cave in De Hoop Nature Reserve and Won- South Africa indicating the distribution ranges of the three species, dergat Cave at the Orange River) as the bats emerged to forage. Only the geographical localities where the respective stimuli classes adult males and nonlactating adult females were kept and imme- were recorded and the two study sites. diately brought to housing facilities in soft cotton bags (a maximum 30 min drive from each of the caves). After experiments, bats were Stimulus Generation released back into their respective caves at night. We observed the bats after releasing them to make sure they showed no signs of The echolocation calls used in this study were previously weakness and could fly off easily. recorded in a standardized way (Odendaal et al., 2014). For all We kept the bats in small mixed-sex groups in a transport cage playback files we chose good quality echolocation calls with a high lined with soft wire mesh (40 25 cm and 25 cm high) with up to signal-to-noise ratio. Because horseshoe bats tune into their resting six bats for up to 3 nights or in a two-person tent (105 105 cm and peak frequency after a period of silence (Schuller & Suga, 1976; 205 cm high) with up to 12 bats for 6 nights. The tent was equipped Siemers, Beedholm, Dietz, Dietz, & Ivanova, 2005) we chose calls with a water bowl and towels to provide perches and hiding places emitted after the first 10 calls in each recording. We decided to use for the bats. The cage/tent was housed in the respective research only calls from males because the RF differs between the sexes in houses inside a quiet room which we darkened and kept at ambient R. capensis (Odendaal & Jacobs, 2011). To minimize effects of temperature and humidity for most of the time but occasionally pseudoreplication, which is the possibility that bats may memorize heated (in March) and cooled (in November). Handling of bats was a set of individuals rather than forming a template of the charac- kept to a minimum; we offered them water twice per day and only teristic call of the population (problem of pseudoreplication; Mc handled them again prior to experiments. We used headlights (at Gregor et al., 1992), we used the highest possible number of in- lowest illumination) when handling bats; otherwise it was dividuals representing each phonetic population and species. We completely dark. We noted the weight of each bat each night to measured six spectral and three temporal parameters of the make sure bats were fed sufficiently. Immediately before experi- echolocation calls (see Audio analysis below and the Results). ments we gave them water and up to 10 mealworms (Tenebrio Parameter measurements of the six playback classes are given in 144 A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154

Table A1 in the Appendix. A discriminant function analysis (DFA) calculated the coefficient of variation (as CV ¼ (SD/mean) 100) to showed that the calls used in the two playbacks from the two test ensure that the selected calls and individuals did not come from the populations, RcaDH85 and RcaLS75, are indeed representative of outer limits of the variability (CV would increase) which would the respective phonetic population (Fig. 2). The DFA was performed increase the possibility of individual signatures in the playback and on factor scores of a principal component analysis to obtain a set of thereby the problem of pseudoreplication (Mc Gregor et al., 1992). uncorrelated acoustic variables. The first four principal components All CVs were lower than the CVs of the published larger samples explained 84% of the variance and represented mainly the fre- (RcaLS75: 1.3% in literature versus 0.8% this study; RcaDH85: 0.8% quency, entropy and duration parameters (PC1: peak frequency of versus 0.5%; RcaHH84: 0.6% versus 0.5%; RcaTF86: 0.8% versus 0.6%; the dominant second harmonic, i.e. RF (0.94); peak frequency of Rcl: 0.9% versus 0.4%; Rda: 1.6% versus 0.9%). the third harmonic (0.87); peak frequency of the first harmonic In an attempt to keep the number of identical call replicates low (0.79); PC2: minimum frequency of final FM component (0.84) we used 5e15 calls for each individual and applied a two-order and entropy of the call (0.75); PC3: duration of call (0.79) and randomization procedure by randomizing first the order of in- distance from start of the call to its maximum energy (disttomax; dividuals and then the order of their calls. To achieve the required 0.78); PC4: intercall interval (0.88)). The stepwise canonical DFA duration of playbacks with the limited number of available good on the four PCs (Wilks' l ¼ 0.00945, approximately F20,296 ¼ 455.4, quality calls per individual we repeated the unique calls randomly P 0.001) indicated that function 1 contributed most to the until we had 20 s long files per test stimulus class and 18.5 min long discrimination between the six playback classes (partial Wilks' habituation files. We inserted each call file with its respective in- l ¼ 0.04), followed by function 2 (0.025), function 3 (0.56) and tervals into the playback. The habituation playback files consisted function 4 (0.77). Accordingly, function 1 is weighted most heavily of 86 unique calls of nine individuals of R. capensis recorded at De by PC1 (coefficient 1.37; RF) and PC3 (coefficient 0.79; call dura- Hoop and 58 unique calls of six individuals of R. capensis from tion and disttomax) and explains 88% of the variance. Function 2 is Lekkersing. We chose four calls (RcaTF86, RcaDH85), five calls marked by PC2 (1.03; minimum frequency of final FM component, (RcaHH84, RcaLS75, Rda85) and six calls (Rcl92) per individual i.e. minimum frequency at the end of the call and call entropy) and depending on the available number of individuals. For the test together both functions explain 99.5%. This increases to 99.9% if playback files, RcaDH85 and RcaLS75, we chose a set of new calls function 3 is added (0.93 marked by PC4 reflecting the parameter not used for the habituation playback file but from the same in- intercall interval). Thus, the most discriminatory power was asso- dividuals used to compile the habituation playback file. The test ciated with the combined effects of peak frequency, call duration, files began with a 23 ms long series of habituation calls (a randomly disttomax, minimum frequency of final FM component and call chosen subset of the habituation calls), followed by the actual test entropy. The overall classification success was 84.5% for the six calls with an averaged interval to the last habituation call. We playback categories (LS: 100%; DH 62%; HH: 85%; TF: 76%; Rcl: decided to precede the test calls with habituation calls to ensure 100%; Rda: 65%). that any dishabituation reaction would be caused by the acoustic The RFs of our playbacks differed slightly from the published properties of the test stimuli and not by the onset of the test values as the available calls used to generate the playbacks shifted playback file because there was a short pause in playback while the mean RF between ±0.1 and ±0.6 kHz. This deviation is well changing files. The series of test calls was followed by a control within the range of natural variability within the populations or stimulus which was a 95 ms white noise and a low-frequency beep species (Jacobs et al., 2013; Odendaal et al., 2014). In addition we for synchronization of video and audio recordings (0.7 kHz, 500 ms) at the same intensity as the calls (Fig. 3). The calls were semisynthesized, meaning that a natural call was 6 RcaDH85 used as a template to create a synthesized copy of it (Avisoft-SASLab RcaHH84 RcaTF86 Pro, v5.2, Avisoft Bioacoustics, Glienicke, Germany) to exclude any RcaLS75 noise or recoding artefacts. All synthesized calls were normalized to Rda85 4 Rcl92 the same intensity level. To estimate appropriate playback intensity Group centroid we recorded two bats inside the box on the perch with a micro- RcaHH84 phone placed next to the speaker through which the playbacks 29 would be played with a medium sensitivity of the microphone RcaTF86 2 36 21 35 24 RcaDH85 (frequency gain on the detector one-third opened). We used the 1928 2 2237 intensity of these recordings to adjust the intensity of the playback Rda85 27 26 output by decreasing the volume of it inside the file until it was 3418 Function 2 8 17 0 1 20 one-third louder than the recorded calls. This increased intensity 16 13 3325 30 1015 9 was necessary to exclude the effect of a sensory bias on the RcaLS75 11 32 23 4 31 6 discriminatory ability of the listeners. The auditory system of 12 ‘ ’ −2 14 Rcl92 horseshoe bats contains an individualized acoustic fovea (Neuweiler, Bruns, & Schuller, 1980; Schuller & Pollak, 1979) which 7 is a narrow filter for increased frequency resolution that spans the 3 individual's RF and covers approximately 8 kHz (Neuweiler, 1990). − 5 4 Frequencies outside the foveal region have threshold levels com- −10 −8 −6 −4 −2 024 6810 parable to other mammals which do not have an acoustic fovea Function 1 (Neuweiler, 1990). Increasing the intensity of our playbacks mini-

Figure 2. Plot of the first two functions of a discriminant function analysis (DFA) based mized the possibility that differences in threshold levels could have on acoustic parameters of the playback stimuli and the experimental individuals. Each resulted in a lack of response by the bats for those playback classes call of each playback class is plotted; numbers indicate the group centroid per indi- containing calls outside their foveal area. We tested the sponta- vidual. Each playback class is colour coded and each species is indicated with different neous response of three bats to the onset of a habituation playback symbols (see key). Individuals grouping with RcaLS75 playback calls on the left-hand to ensure that the bats' reaction towards the increased intensity of side (4to8 on the x axis) are test subjects from the study site Lekkersing, RcaLS75, whereas all other individuals on the right-hand side (1toþ4 on the x axis) were from the playbacks was not a startle response (fast body contraction, De Hoop, e.g. RcaDH85. cringe). All files began with a linear fade-in from 0 dB to maximum A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 145

Habituation Dishabituation HabSTART HabEND TEST MOT On Off PGH Gap SYNC

15 s 15 s 15 s Time

Figure 3. Schematic overview of the different phases of the habituation e dishabituation experiment. The habituation phase has a variable duration whereas the duration of the dishabituation phase is fixed to 15 s. HabSTART contained the first calls of the habituation phase and HabEND the final calls. HabEND contained a brief moment of silence (gap) between two series of habituation calls (last series is postgap habituation, PGH). The third analysed time section is the 15 s long TEST which contained the test playback calls. At the end of the TEST calls a motivational stimulus (MOT) was played back (white noise) followed by a synchronization beep (SYNC). amplitude within the first call to avoid the clicking noise emitted by When the bat habituated to the habituation stimulus, i.e. showed the loudspeaker when playback files began with maximum no visible movements for 20 s, we played back the test stimulus. amplitude. The fading-in and -out was also used for the control and Again, the response to the test and motivational control stimulus synchronization sounds (one-quarter of each element). was noted (reacted, did not react). After each trial we left the bat for at least 30 s in the box before it was brought back to a separate Experimental Set-up room, where it was fed and later released into the housing quarters. If any execution errors occurred during playbacks, the respective Experiments took place in a custom-made experimental box trial was repeated as the last trial for that bat. Trials were also measuring 77.5 38.0 cm and 38.0 cm high (Genesisacoustics, Port repeated if a bat did not show a response to the habituation onset to Elizabeth, South Africa) similar to that used by Schuchmann and prevent subsequent false negative responses to the test stimuli. In Siemers (2010) which was set up in a separate room. It was lined this case we stopped the habituation playback and started it again with 10 cm thick, high-frequency, sound-absorbing foam covered after 30 s. If a bat did not react to the dishabituation stimulus and with glass cloth. It contained a perch for the bat on one end and a the motivational control played back at the end of each dis- loudspeaker (ScanSpeak, Avisoft Bioacoustics) at the opposite end. habituation stimulus we repeated the trial (again to control for false A camcorder (DCR-SR42E Sony Corporation, Minato, Tokyo, Japan) negatives). At both study sites we ran one to two trails per bat per with activated night-shot function was placed in one of the corners night to shorten the period of captivity for each. Seven of a total of below the loudspeaker and pointing at the perch. An ultrasound 32 bats listened to three trials in 1 night because of technical errors detector (D1000X, Pettersson Elektronik, Uppsala, Sweden) with a during the previous night. We ensured that playback classes were calibrated microphone was placed in the other corner below the used in a balanced design across nights and individual bats that did loudspeaker and pointed towards the perch. The loudspeaker was multiple trials were used at different times of the night. At the connected to an ultrasonic power amplifier (Avisoft Bioacoustics) study site in the northwest, LS, we limited the playback set to five and via a high-speed sound card (DAQCard-6062E, National In- trials because of the extremely high temperatures to reduce the struments, Austin, TX, U.S.A.) to a laptop (Hewlett and Packard risks of bats succumbing to heat exhaustion. Here we excluded the Mobile Workstation, Palo Alto, CA, U.S.A. with the Microsoft Win- playbacks RcaTF86 and RcaHH84 and added the trial using dows XP Professional, SP3, Microsoft Corporation, Redmond, CA, RcaDH85calls (allopatric conspecific) for habituation and Rda85 U.S.A. operating system). The camcorder was connected to a calls (sympatric heterospecific) as the test stimulus to test whether monitor outside the box to allow observation of the behaviour of bats can discriminate a different phonetic population of the same the bat inside during experiments. To control the correct broad- species from a different sympatric species using a very similar casting of the playbacks we used Avisoft recorder software (rec-ni resting peak frequency. To facilitate direct comparison, the exper- v4.2.15, Avisoft Bioacoustics). All settings on the equipment that iments were performed by the same person at each study locality. could influence the output characteristics of the playbacks or the characteristics of the audio recordings were kept constant (ampli- Recording of Experiments fier volume, frequency gain of the bat detector) as well as the equipment components themselves. Each trial was synchronously recorded on video and audio from the moment each bat was placed on the perch inside the box until Experimental Procedure we took it out of the box at the end of the trial. Audio and video recordings were synchronized by using the low-frequency beep We used the following test regime: each bat listened to all played back at the end of each test playback, SYNC in Fig. 3, which stimuli classes after being habituated to calls from its own popu- was recorded on both the audio line of the camcorder and the audio lation on consecutive nights. The order of testing the bats was file of the bat detector to serve as the reference point between randomized and we ensured that each bat was not tested at the video and audio. same time each night. The order of test stimuli was also random- Audio recordings were done with a 385 kHz sample rate, 16 bits ized using the BoxeBehnken Design, available in Statistica (v6.1, depth and stored on internal CF cards of the bat detector as StatSoft Inc., Tulsa, OK, U.S.A.). The maximum duration for a trial Waveform Audio File files (.wav). Large audio files were later split was set to 40 min which was occasionally reached by individual (Wave Splitter, v2.10, ClaudioSoft software, Ile-de-France, France) to bats on the first 2 nights. allow analyses in the sound analysis software (Avisoft-SASLab Pro, The experiments began at around sunset. Each bat was brought v5.2, Avisoft Bioacoustics; BatSound Pro v3.31, Pettersson Elek- to the experimental box in a small cotton bag and placed on the tronik AB). perch. If the bat crawled or flew off the perch we placed it back onto Video recordings were done using the internal infrared light of the perch until it stayed there. The playback started when the bat the camcorder (DCR-SR42E Sony Corporation) with the night shot hung calmly at the perch and showed no visible movements for at function activated providing sufficient illumination for high-quality least 20 s. The habituation stimulus was played back and the im- videos stored on the internal hard drive of the camcorder. Videos mediate response of the bat was noted (reacted, did not react). were transferred from the camcorder onto the laptop using the 146 A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 camcorder's software (Picture Motion Browser v2.0.06 Sony Cor- switching between playbacks interrupted the habituated state. We poration). Videos had to be converted (v3.0.057, Mangold Video verified that bats were still habituated by comparing the duration Converter Pro, Mangold International, Arnstorf, Germany) to the of attentive behaviours during HabEND versus the last series of Audio Video Interleave (.avi) format (video: AVC format, 8 bits postgap habituation calls (PGH, see Fig. 3) using Wilcoxon depth, 25 fps frame rate, 265 Kbps bit rate; audio: PCM format, matched-pairs tests. We controlled for false positives by using the 16 bits depth, 48 kHz sampling rate, 1536 Kbps bit rate) to ensure control test playback, which contained calls of the same population correct playback in the behavioural analysis software (InterAct v8.0, used in the habituation file, to assess the frequency of occurrence of Mangold International). spontaneous recoveries from habituation, i.e. a false positive response, We controlled for false negatives, i.e. the failure of the bat Analyses of Experiments to react to a test stimulus because of sensory/experimental fatigue (Pancratz & Cohen, 1970), by determining whether the bat General statistics methods responded to the motivational control. We tested whether bats The global level of significance was 5% but we calculated the were habituated before playing back the test stimulus by Bonferroni-corrected P values to account for multiple comparisons. comparing the occurrence of behaviours at the beginning and end Full details of statistical analyses are given with each test, where of the habituation phase as well as the duration of attentive be- necessary, in the Methods below and in the Results. All statistical haviours. Each trial was tested using chi-square tests in which the tests were two tailed and carried out in Statistica v6.1 and SPSS v22 expected number of frames for ‘inactive/calm’ and ‘attentive’ for (IBM SPSS Statistics for Windows, IBM Corporation, Armonk, NY, HabEND was the number of frames counted in the HabSTART phase U.S.A.). and compared to the observed number of frames counted during HabEND. The duration of attentive behaviours during HabSTART Video analysis and HabEND was tested across all trials using a Wilcoxon matched- The first field trip in March 2012 served as a pilot study to train pairs test. The effects of long-term habituation on the general level the experimenter and assistants and to test the number of possible of responsiveness of the bats needed to be evaluated because the playback trials per individual. In addition, we obtained a set of a experiment was done over several days. This was done by priori behaviours. We focused on behaviours that are informative in comparing the duration of attentive behaviours for the beginning of the detection of differences in the response of the bats (only slightly the habituation playback across the experimental nights using a modified during video analysis if a bat showed a new behaviour). Friedman ANOVA. After confirming the validity of the trials we These behaviours allowed us to quantify (occurrence, count data) tested discrimination between the bat's own population echolo- and qualify (duration) each subject's response to the stimuli. The cation calls versus the different classes of test playback echoloca- behavioural variables (listed and defined in Table A2 in the Ap- tion calls by comparing the duration of attentive behaviours during pendix) were classified a posteriori into the main categories inac- the end of the habituation phase (HabEND) and during test play- tive/calm, active and attentive behaviours. Attentive behaviours are backs (TEST). We did this comparison for each playback category a subset of active behaviours. We defined attentive behaviours as across individuals using Wilcoxon matched-pairs test. behaviours that imply a stimulus-directed response of the bat to the playbacks by either orienting itself towards the sound source Audio analysis (e.g. head raised), by revealing a listening response (e.g. ears Large audio files were split using Wave Splitter v2.10 to obtain a twitching) or by showing a startle response (cringing as indicated manageable file size. Corrupted wav files were recovered using by leg contraction). Active behaviours, on the other hand, which are Audacity v2.0.3 (Audacity Team, SourceForge.net). rated nonattentive comprise movements of the bat that occur when Acoustic measurements (Fig. 4) were taken on calls of each bat it settles down (e.g. scrambling) and do not indicate any reaction to and of the playback stimuli using the automatic measurement the acoustic stimulus (e.g. wing stretching). This classification was function in Avisoft SASLab Pro v5.2. We analysed 10e30 calls of verified throughout the video analysis and categorized by three each experimental individual from recordings obtained during different observers. experiments. We chose calls of similarly good quality at the Each trial was analysed frame by frame with 25 frames/s using beginning of the trial before any playback was started and the bats the software InterAct (InterAct 8.0, Mangold International). We showed orientation behaviour in the box, such as scanning the restricted the analysis to three 15 s time periods (Fig. 3): the first environment. We, again, chose calls emitted after the first 10 calls, 15 s of the habituation (HabSTART), the last 15 s of the habituation avoiding lower frequency calls of the tuning in phase (the first se- (HabEND) and the 15 s long test stimuli (TEST) ending with the ries of calls emitted after a silent period show lower RFs and motivational control stimulus (MOT) and the synchronization gradually ‘tuned in’ to the individuals' typical RFs; Siemers et al., signal (SYNC). 2005). Two observers (A.B. and D.S.J.) coded the behaviours of bats in We measured the duration of each call, the interval between the the videos of the trials. In all cases, the researchers had no onsets of two calls, the distance from the beginning of each call knowledge about the stimuli class used in the videos. To ensure until it reached its maximum energy, the peak frequency of each consistency in video coding, 10 randomly chosen trials were double call, the frequency of the first and third harmonics, the minimum coded by both researchers and tested for interobserver reliability. frequency of the frequency-modulated components at the begin- We used Cohen's kappa for count data (Interact v8.0 electronic ning and at the end of the prominent second harmonic and the manual, Mangold International) and Kendall's tau to test the du- call's entropy which quantifies the pureness of sounds (its value is rations of the behaviours for congruence between the observers. zero for pure-tone signals and one for random noise; see Avisoft- Finally, we tested whether females and males differed in their SASLab Pro manual v5.1, p. 162). response behaviour by comparing the duration of active behaviours with a ManneWhitney U test. RESULTS Trials were regarded as invalid and thus not analysed if the data for an individual were incomplete, if a bat did not react to the initial In total we obtained complete and analysable data for 36 bats: habituation playback, if the bat left the perch during the final phase 15 (seven males; eight females) were from Lekkersing and 21 (10 of habituation and/or the test playback or if the pause caused by males; 11 females) from De Hoop. A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 147

Oscillogram: The bats from De Hoop, RcaDH85, reacted in 85 of 126 trials 100 (67.5%) with active behaviours. Twice the bats reacted with non- attentive active behaviours to the control playback RcaDH85 (2) (Fig. 5b). In trials with test playback calls of the sympatric 0 R. clivosus, Rcl92, all bats reacted and in all but one trial bats reacted (1) when we played the allopatric R. damarensis, Rda85, and RcaLS75 echolocation calls. When hearing echolocation calls from conspe- fi 50 ci cs using slightly different resting peak frequencies, RcaHH84 −

Relative intensity (%) 100 and RcaTF86, the bats reacted ambiguously: 10 reacted to calls from Time (ms) Sonagram: Power spectrum: RcaHH84 bats and 12 to calls from RcaTF86 bats. Here, as with the (6) RcaLS75 bats, we found a significant difference between the re- 120 * sponses to the different playback classes (Cochran Q test: Q5 ¼ 53.808, N ¼ 21, P < 0.001). (8) (5) The bats from Lekkersing, RcaLS75, showed a dishabituation * 80 * response to the playbacks of RcaDH85, RcaTF86 and the sympatric * (3) species Rda85 when habituated to their own population calls (7) RcaLS75 (Wilcoxon matched-pairs test: T ¼ 0.00, Z ¼ 0.41, N ¼ 15, P ¼ 0.001; T ¼ 0.00, Z ¼ 3.42, P ¼ 0.001; T ¼ 19.0, Z ¼ 0.85, N ¼ 15, 40 Frequency (kHz) (4)

Frequency (kHz) * P ¼ 0.004; Fig. 6a). However, when previously habituated to calls from the other phonetic population, RcaDH85, they showed no 10 attentive reaction to the calls of the sympatric species Rda85. They 100 0 also showed no significant reaction to calls from their own popu- Time (ms) Intensity (dB) Time (ms) lation (i.e. no false positive; Fig. 6a).

Figure 4. Example of two R. capensis echolocation calls. The parameters measured for When habituated with their own calls, the bats from De Hoop in the acoustic analysis are indicated in the oscillogram, the sonagram or the power the south, RcaDH85, mostly did not discriminate between calls spectrum. Lines indicate duration measurements, asterisks indicate points of mea- from acoustically similar phonetic populations of their own species, surements and numbers next to asterisks refer to each parameter: (1) call duration, (2) RcaHH84 and RcaTF86, but did show a significant response to the interval, (3) distance to maximum amplitude, (4) frequency fundamental harmonic, (5) acoustically dissimilar population RcaLS75 and the other two spe- peak frequency second harmonic (RF), (6) frequency third harmonic, (7) minimum frequency at start (second harmonic), (8) minimum frequency at end (second cies Rda85 and Rcl92 although only one of them, Rcl92, co-occurs harmonic). with them at De Hoop (Wilcoxon matched-pairs test: T ¼ 1.0, Z ¼ 3.88, N ¼ 21, P < 0.001; T ¼ 4.0, Z ¼ 3.77, N ¼ 21, P < 0.001; The interobserver reliability test indicated high conformity be- T ¼ 0.0, Z ¼ 4.02, N ¼ 21, P < 0.001; Fig. 6b). They did not react to fi k > ¼ tween the observers (Cohen's kappa coef cient: 0.77, N 10, the playbacks of the control calls of their own population (i.e. no < P 0.05) as well as concordance of duration of behaviours (Kendal's false positive; Fig. 6b). fi t ¼ ¼ < tau rank correlation coef cient: 0.7, Z 3.580, P 0.001). We The difference in discriminatory ability between the two test found no difference in the response behaviour between males and populations is also apparent when comparing the response females and therefore grouped all bats for each sample site (Man- strength in relation to the acoustic similarity of the playbacks e ¼ ¼ n Whitney U test: RcaLS75 bats: U 625.0, Zadjusted 0.78, (Fig. 7). The playback from the phonetic population RcaLS75 is ¼ ¼ ¼ ¼ Nmales 7, Nfemales 8, P 0.431; RcaDH85 bats: U 1209.5, acoustically dissimilar to the other playbacks based on the first two ¼ ¼ ¼ ¼ Z 1.32, Nmale 10, Nfemale 11, P 0.188). We observed a mod- functions of a DFA which predominantly consisted of frequency and fi erate and nonsigni cant decline in responsiveness over experi- temporal parameters (see Fig. 2 and section Stimulus Generation in c2 ¼ mental days (Friedman ANOVA: RcaLK75 data set: 14 1.200, the Methods for details). Bats from this population showed strong ¼ c2 ¼ ¼ P 0.549; RcaDH85 data set: 20 10.987, P 0.052). Our tests responses towards all playbacks from other species or conspecific fi con rmed a successful habituation at the end of the habituation phonetic populations (red, green and grey columns on the left in c2 phase for each trial (chi-square test: 1 3.84, P 0.05, single Fig. 7a) and low response levels towards its own calls (control, grey tests not shown). The duration of attentive behaviours also column on the right in Fig. 7a). Accordingly, the bats from the fi con rmed a decrease in active behaviours from the beginning to phonetic population RcaDH85 showed a strong response towards the end of the habituation phase across individuals per playback the acoustically dissimilar playback of the conspecific phonetic class (see next section; Wilcoxon matched-pairs test: RcaLS75 data population RcaLS75 (grey column on the right in Fig. 7a), but an ¼ ¼ < ¼ set: T 39.50, N 75, P 0.01; RcaDH85 data set: T 84.50, almost similarly strong response to the acoustically similar play- ¼ < N 126, P 0.001). back of the allopatric heterospecific Rda85 (green column in Fig. 7a). Correspondingly, the relationship between acoustic dis- Discrimination Task tance and response strength was stronger for the bats from Lek- kersing RcaLS75 (Fig. 7b) and weaker for the bats from De Hoop The bats from Lekkersing, RcaLS75, showed a reaction by RcaDH85 (Fig. 7c). Although the acoustic distance of Rda85 calls to becoming active again in 51 of 75 trials (68%). In trials with the test echolocation calls of RcaDH85 was low, the response strength of playback RcaDH85, RcaTF86 or the sympatric Rda85, all 15 bats RcaDH85 bats was high towards the playbacks of Rda85 (greyish showed a reaction in all trials (Fig. 5a). In one trial of the control area in the graphs). The respective narrow area of acoustic distance playback RcaLS75 one bat showed a reaction (false positive). In this (as set by the variation of calls used by conspecifics of the listening case the reaction was a nonattentive active behaviour. When bats) for the RcaLS75 bats shows lower response levels towards habituated with the calls of the other phonetic populations, playback calls inside this area. The increased response strength RcaDH85, and tested against the sympatric Rda85, bats showed a showed that bats from De Hoop discriminated between calls from reaction in five of 15 trials. Overall, this is a significant difference their own population (RcaDH85) and calls from the allopatric between the playback classes (Cochran's Q test: Q4 ¼ 46.359, heterospecific (Rda85), despite overlapping frequencies and gen- N ¼ 15, P < 0.001). eral acoustic similarity. In contrast, bats from Lekkersing showed no 148 A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154

17 (a) increased response, i.e. no discrimination between these two similar echolocation playbacks (RcaDH85 and Rda85) compared to 15 100% 100%7% 100% 33% the control (own population calls). 13 11 DISCUSSION 9 7 In both test populations, R. capensis bats were able to discrimi- 5 nate between their own calls and calls of the respective sympatric YES horseshoe bat species. Furthermore, the echolocation calls of bats 3 NO from Lekkersing were perceived as distinctly different from those of 1 all other phonetic populations of R. capensis by the bats from both Hab.: own75 own75 own75 own75 DH85 phonetic populations. However, our results suggest that the bats Test: DH85 TF86 own75 sympat85 sympat85 from the two test populations differed in their ability to discrimi- nate between calls of conspecifics and heterospecifics of the same Playback combination for RcaLS75 bats resting peak frequency. De Hoop bats displayed this ability but 23 (b) Lekkersing bats did not. The prediction of the ACH that increased signal overlap between calls of sympatric species should increase Number of trials 21 9% 52% 43% 95% 100% 95% the rates of misclassification of calls from conspecifics or hetero- 19 specifics was therefore confirmed for bats from Lekkersing. How- 17 ever, bats from De Hoop were able to discriminate between species using overlapping RFs. This confirms that bats are able to use 15 additional acoustic cues besides frequency to discriminate con- 13 specifics from heterospecifics. 11 This study also provides evidence that overlapping signals 9 among species occurring in the same habitat can be accommodated by the evolution of increased perceptual acuity in the receiver. 7 Although this has been shown in other taxa (e.g. birds: Seddon & 5 Tobias, 2010; Tobias et al., 2014; frogs: Amezquita et al., 2006; 3 2011; Marshall et al., 2006; insects: Nosil et al., 2003; Jang & 1 Gerhardt 2006) to our knowledge this is the first time it has been fi Hab.: own85 own85 own85 own85 own85 own85 shown in a mammal. It is also the rst observation in any mammal of the evolution of differences in perceptual acuity within the same Test: own85 HH84 TF86 LS75 sympat92 allopat85 species mediated by differences in the complexity of signalling Playback combination for RcaDH85 bats assemblages and constraints imposed by the dual function of the

Figure 5. Number of trials in which the bats from the two study sites (a) LS and (b) DH trait (see below). This observation suggests that the availability of discriminated (yes) or did not discriminate (no) the calls of the test playback from the signal space for a particular species may mediate the evolutionary calls in the preceding habituation playback by showing activity. response of this species to competition for a unique communication channel from sympatric species. Signal space is the range of

50 (a) (b) 45 *0.001 *0.003 *0.001 *0.002 *0.001 *0.001 *0.001 *0.001 *0.001 *0.001 *0.001 40 *0.001 *0.001 0.5 *0.005 0.5 0.5 0.05 (*) 0.04 *0.001 *0.001 *0.001 35 30 25 20 15 10

5 Duration of attentive behaviour (s) behaviour attentive of Duration

0

TEST TEST TEST TEST TEST

HabEND HabEND HabEND HabEND HabEND

HabSTART HabSTART HabSTART HabSTART HabSTART

HabSTART HabSTART HabSTART HabSTART HabSTART HabSTART

HabEND HabEND HabEND HabEND HabEND HabEND

TEST TEST TEST TEST TEST TEST Hab.: own75 own75 own75 own75 DH85 own85 own85 own85 own85 own85 own85 Test: DH85 TF86 own75 sympat85 sympat85 own85 HH84 TF86 LS75 sympat92 allopat85 Playback combination for RcaLS75 bats Playback combination for RcaDH85 bats

Figure 6. Duration of attentive behaviour per study group (a) RcaLS75 and (b) RcaDH85, per playback class and per analysed time section (HabSTART white box, HabEND grey box, TEST black box). P values of each Wilcoxon matched-pairs test are given above each pairwise comparison whereas an asterisk indicates a significant result for a ¼ 0.05 and abonferroni ¼ 0.02. The box plots show median (horizontal line), 25th and 75th quartiles (box), nonoutlier range values falling within 1.5 times the interquartile range (whiskers), outliers falling more than 1.5 times the interquartile range from the edge of the box (circles) and extremes falling more than 2.0 times the interquartile range from the edge of the box (crosses). A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 149

(b) 16 (a) 3

14 2 RcaTF86 RcaLS75

12 Rda85 1

10 0

RcaDH85 −1 8 040 80 120 180

(c) 6 3

Response strength

Mean duration of response (s) 4 2

2 1

* 0 0 2 1 7 8 6 −1 0 4 5 (’dur, disttomax’)−1 2 3 040 80 120 160 −2 1 −1 0 Acoustic distance RcaDH85 Rda85 RcaDH listeners RcaHH85 Rcl92 RcaTF86 RcaDH85-RcaDH85 RcaLS listeners RcaLS75

Figure 7. Relationship between the strength of response of bats from two phonetic populations of R. capensis and acoustic similarity/dissimilarity of the playback classes. Response strength was measured as duration of attentive behaviours occurring as a response to the dishabituation playback class. (a) Three-dimensional representation of the relationship between response strength and acoustic space. The response strength is represented as the vertical axis and the acoustic space is demarcated by function 1 and function 2 of a discriminant function analysis on the calls of the playbacks (indicated above each coloured response column). Playbacks with which both populations were tested are depicted to directly compare their response strength. Thus, for each playback class two response columns exist representing the mean duration of attentive behaviours of the two test populations RcaDH85 and RcaLS75. The two grey columns represent the control playback for RcaLS (column on the right) and for RcaDH (second from the left). The red column on the far left shows the response to the playback of a different phonetic population, RcaTF86, and the centre green column (second from the right) the responses towards the heterospecific species R. damarensis, Rda85 (allopatric for RcaDH85, sympatric for RcaLS75). The asterisk (*) next to this column shows the response of RcaLS75 bats towards the stimulus combination RcaDH85 versus Rda85 (see Methods for details). (b, c) Standardized strength of response of (b) RcaLS75 and (c) RcaDH86 bats to the different playback classes as a function of acoustic distance between habituation and test playback for the two phonetic populations. Linear regression lines and 95% confidence intervals are shown 2 2 (RcaLS75: R ¼ 0.296, F1,58 ¼ 25.840, P 0.001; RcaDH85: R ¼ 0.174, F1,124 ¼ 26.185, P 0.001. Each playback class is colour coded and each species is indicated with different symbols (see key below graph). acoustic signals theoretically available to a species into which its is more broadly applicable to other taxa. We suggest the following signal could diverge and is delimited by sympatric species but also formulation of the hypothesis. In an assemblage of sympatric het- by constraints imposed by the dual function of a sensory system. erospecific animals using acoustic signals, multidimensional acoustic Thus, where signal space permits, potential misclassification of space is partitioned in such a way that each species occupies a signals as a result of overlapping signals used by sympatric species distinct acoustic space facilitating intraspecific communication and may be minimized through the evolution of signal divergence species discrimination. This would also distinguish the ACH from the through character displacement. However, where signal space is spectral partitioning hypothesis (Amezquita et al., 2011). limited the evolution of increased perceptual acuity among re- ceivers may allow both coexistence and convergence of signals (see Local Acoustic Signalling Assemblages and Perceptual Acuity also Seddon & Tobias, 2010). The use of multiple cues in acoustic signals for vocal recognition of The fact that the bats from the two study sites differed in their kin, mates or competitors is not a novel discovery and is well ability to use additional acoustic cues in echolocation calls might be established for birds, anurans and mammals (mammals: a result of differences in the local acoustic assemblages. Lekkersing Hammerschmidt & Todt,1995; anurans: Vignal & Kelley, 2007; birds: is remote and bats there are isolated from other R. capensis pop- Vignal, Mathevon, & Mottin, 2008) although species discrimination ulations as evidenced by limited historical gene flow between based on multiple acoustic cues in echolocating animals was only Lekkersing and other populations (Odendaal et al., 2014). Rhinolo- suspected for bats (Schuchmann et al., 2012). The ACH, as currently phus capensis at Lekkersing are therefore only exposed to the calls applied to bats (e.g. Heller & von Helversen,1989; Jacobs et al., 2007), of a single different rhinolophid species and were unable to is thus too narrow in its formulation as it is based on resting fre- discriminate between conspecific and heterospecific quency alone. It should be expanded to incorporate multiple call (R. damarensis) calls that were similar but different to its own. This components as was initially done for anurans (Duellman & Pyles, means that they may use a simple rule of thumb that bats calling at 1983) so that it more accurately describes the situation in bats and different frequencies are heterospecifics. This rule of thumb results 150 A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 in their being unable to distinguish between heterospecifics and lower frequencies. Avoiding call overlap with R. clivosus by using conspecifics echolocating at similar and higher frequencies. In higher frequency calls would probably also require a reduction in contrast, the acoustic assemblage for bats from De Hoop is more body size for increased flight manoeuvrability (Norberg & Rayner, complex than that encountered by R. capensis at Lekkersing. At De 1987) to deal with the decreased detection range resulting from Hoop and the region around it there are populations of bats of the increased atmospheric attenuation of higher frequency calls same species with slightly different RFs in close geographical (Lawrence & Simmons, 1982; Schnitzler & Kalko, 2001). All rhino- proximity. These bats are thus likely to be exposed to a higher lophids echolocating at higher frequencies than R. clivosus (e.g. variation of frequencies within their own species and the frequent Rhinolophus swinnyi and Rhinolophus denti) are much smaller than gene flow between these populations in the past (Odendaal et al., R. capensis (Monadjem et al., 2010). At De Hoop there may thus be a 2014) suggests an ability to identify individuals from these pho- trade-off between the orientation/food acquisition functions of netic populations as conspecifics. The acoustic challenge for these echolocation and its communicative role, i.e. selection to optimize bats is further complicated by the co-occurrence of the hetero- these functions is exerted in opposite directions. The costs to the specific R. clivosus which has a frequency range overlapping, albeit orientation and food acquisition functions of echolocation may slightly, with that of its own. Finer levels of species discrimination have precluded the evolution of call frequencies in R. capensis that and conspecific recognition based on frequency, as well as other were divergent from those of R. clivosus. Instead, communication acoustic cues coded in echolocation calls, may therefore provide a via echolocation at De Hoop may have been optimized through the selective advantage. This interpretation of our data is further sup- evolution of finer levels of discrimination based on other compo- ported by the ability of R. capensis at De Hoop to clearly discrimi- nents of the call. This is similar to the evolution of overlapping male nate between calls of their own species and those from an allopatric songs across several congeneric species in a complex multispecies heterospecific, R. damarensis, using the same resting frequency. We community of forest-dwelling birds (Tobias et al., 2014; Tobias & therefore propose that adaptation of the receiver's sensory systems Seddon, 2009). This overlap was apparently driven by the dual to the local acoustic assemblage may have resulted in a hierarchy of function of birdsong in mate attraction and interspecific territory discrimination abilities within the same species. Simple rules of defence in shared habitats. Although mate attraction requires thumb based on divergent call frequencies alone may provide unique, individualized male song, territory defence selects for sufficient selective advantage in a simple acoustic assemblage with convergence in male song among different species allowing them more available signal space, such as that experienced by R. capensis to function well in advertising the male's territory across species. at Lekkersing, but additional acoustic cues to discriminate species Overlapping male songs have, however, resulted in the evolution of are necessary when bats live in a more complex acoustic assem- increased perceptual acuity among female receivers allowing the blage with less available signal space. continued functioning of birdsong in intraspecific communication (Tobias et al., 2014). In this example increased perceptual acuity is Perceptual Acuity and Evolutionary Constraints driven by constraints imposed by the dual function of birdsong in communication whereas in our study it is driven by constraints Communication signals, especially those with dual functions, imposed by selection for noncommunicative functions, i.e. orien- may be precluded from reaching their optimum character states as tation and food acquisition. The validity of the ACH as an expla- a result of evolutionary constraints. The primary functions of nation for call frequency divergence within a species is therefore echolocation are orientation and food acquisition. Encoding unique tempered by the different selection pressures within the local communication cues in echolocation may have an adverse impact acoustic assemblage that each population may experience. The on these functions. However, an information-carrying signal that is model system of R. capensis thus provides a novel example of the produced almost constantly and is freely available is a likely evolution of mammalian communication systems by revealing the candidate to be used for communication. Signals emitted by an receiver's adaptation to evolve or maintain species-specific animal, voluntarily or involuntarily, provide useful information, for communication. example about feeding grounds and suitable roosts (Barclay, 1982; Ruczynski, Kalko, & Siemers, 2007), the presence of mates (Behr & Conclusions Von Helversen, 2004; Leippert, 1994), prey (Page, Ryan, & Bernal, 2013; Siemers, Kriner, Kaipf, Simon, & Greif, 2012), a threat Species discrimination ability, even within the same species, is (Mariappan, Bogdanowicz, Marimuthu, & Rajan, 2013; Russ, Jones, dependent on local acoustic signalling assemblages that select for Mackie, & Racey, 2004) or attributes of the sender (Fenton, 2003; the ability to discriminate and may involve a trade-off between the Siemers et al., 2005) and its behavioural state (Bastian & Schmidt, different functions of a multifunctional trait. Acoustic species 2008). Nevertheless, the communicative function of echolocation discrimination can be achieved by the partitioning of signal space can only arise or be fine-tuned if it does not compromise the pri- via divergence of the sender's calls or through increased perception mary functions of echolocation which are fundamental to the bat's abilities of receivers. survival. This might explain why there appears to be so little par- titioning of sonar frequency bands at De Hoop, contrary to the ACH, Acknowledgments whereas at Lekkersing the anomalously low calls of R. capensis may have evolved because it avoids overlap with R. damarensis.At We specifically thank Lizelle J. Odendaal for her contribution in Lekkersing selection that optimizes echolocation for orientation collecting echolocation calls of R. capensis, for contributing to the and food acquisition and selection that allows more effective preparation of playback files, and for partially doing the experi- communication operate in the same direction, i.e. lowering call ments with us. We also thank Bjorn€ Siemers and Robert Barclay for frequency. Lower frequency calls increase prey detection distances crucial comments on the experimental design used in this study in the relatively open habitat at Lekkersing (Odendaal et al., 2014) and practical advice during our first field trip. Echolocation data and avoid overlap with R. damarensis. In contrast, the dense vege- were collected by Hassan Babiker, Megan Cunnama and Orsilla tation within which these bats forage at De Hoop probably selects Schmit. We had great help in capturing and caring for bats during for higher frequencies (Odendaal et al., 2014) whereas advantages the experiments from Tinyiko Maluleke, Gregory Mutumi and that accrue through more effective communication as a result of Robert Raw. We thank the editors and two anonymous referees for nonoverlapping call frequencies with R. clivosus would select for their constructive comments, which helped us to improve the A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154 151 manuscript. This study was financially funded by research grants to Hammerschmidt, K., & Todt, D. (1995). Individual differences in vocalisations of D.S.J. from the South African Research Chair Initiative of the young barbary macaques (Macaca sylvanus): a multi-parametric analysis to identify critical cues in acoustic signalling. Behaviour, 132(5/6), 381e399. Department of Science and Technology and administered by the Heller, K.-G., & von Helversen, O. (1989). Resource partitioning of sonar frequency National Research Foundation (NRF) of South Africa (GUN 64798), bands in rhinolophoid bats. Oecologia, 80(2), 178e186. as well as by postdoctoral research grants to A.B. (URC Postdoctoral Jacobs, D. S. (n.d.) [Audio recordings of juvenile Rhinolophus clivosus.] Unpublished raw data. research grant of UCT (URC-BSTANN006) and NRF-freestanding Jacobs, D. S., Babiker, H., Bastian, A., Kearney, T., van Eeden, R., & Bishop, J. M. (2013). postdoctoral fellowship for Non-South Africans (88359)). Phenotypic convergence in genetically distinct lineages of a Rhinolophus species complex (Mammalia, Chiroptera). PLoS One, 8(12), e82614. Jacobs, D. S., Barclay, R. M. R., & Walker, M. (2007). The allometry of echolocation References call frequencies of insectivorous bats: why do some species deviate from the pattern? Oecologia, 152, 583e594. Altringham, J. D., & Fenton, M. B. (2003). Sensory ecology and communication in the Jang, Y., & Gerhardt, H. C. (2006). Divergence in female calling song discrimination Chiroptera. In T. H. Kunz, & M. B. Fenton (Eds.), Bat ecology (pp. 90e118). Chi- between sympatric and allopatric populations of the southern wood cricket caco, IL: The University of Chicago Press. Gryllus fultoni (Orthoptera: Gryllidae). Behavioral Ecology and Sociobiology, Amezquita, A., Hodl,€ W., Lima, A. P., Castellanos, L., Erdtmann, L., & De Araújo, M. C. 60(2), 150e158. (2006). Masking Interference and the evolution of the acoustic communication Jones, G. (1997). Acoustic signals and speciation: the roles of natural selection and system in the Amazonian dendrobatid frog Allobates femoralis. Evolution, 60(9), sexual selection in the evolution of cryptic species. Advances in the Study of 1874e1887. Behavior, 26, 343e344. Amezquita, A., Flechas, S. V., Lima, A. P., Gasser, H., & Hodl,€ W. (2011). Acoustic Jones, G. (1999). Scaling of echolocation call parameters in bats. Journal of Experi- interference and recognition space within a complex assemblage of den- mental Biology, 202,3359e3366. drobatid frogs. Proceedings of the National Academy of Sciences of the United Jones, G., & Siemers, B. M. (2010). The communicative potential of bat echolocation States of America, 108(41), 17058e17063. pulses. Journal of Comparative Physiology A, 197(5), 447e457. Anderson, M. J., Ambrose, L., Bearder, S. K., Dixon, A. F., & Pullen, S. (2000). Intra- Jones, G., & Van Parijs, S. M. (1993). Bimodal echolocation in pipistrelle bats: are specific variation in the vocalizations and hand pad morphology of southern cryptic species present? Proceedings of the Royal Society B: Biological Sciences, lesser bush babies (Galago moholi): a comparison with G. senegalensis. Inter- 251,119e125. national Journal of Primatology, 21(3), 537e555. Kingston, T., Jones, G., Zubaid, A., & Kunz, T. H. (2000). Resource partitioning in Barclay, R. M. R. (1982). Interindividual use of echolocation calls: eavesdropping by rhinolophoid bats revisited. Oecologia, 124(3), 332e342. bats. Behavioral Ecology and Sociobiology, 10(4), 271e275. Kingston, T., Lara, M. C., Jones, G., Akbar, Z., Kunz, T. H., & Schneider, C. J. (2001). Bastian, A., & Schmidt, S. (2008). Affect cues in vocalizations of the bat, Megaderma Acoustic divergence in two cryptic Hipposideros species: a role for social se- lyra, during agonistic interactions. Journal of the Acoustical Society of America, lection? Proceedings of the Royal Society B: Biological Sciences, 268(1474), 124(1), 598e608. 1381e1386. Behr, O., & Von Helversen, O. (2004). Bat serenades-complex courtship songs of the Kingston, T., & Rossiter, S. J. (2004). Harmonic-hopping in Wallacea's bats. Nature, sac-winged bat (Saccopterix bilineata). Behavioral Ecology and Sociobiology, 56, 429(6992), 654e657. € 106e115. Knornschild, M., Jung, K., Nagy, M., Metz, M., & Kalko, E. (2012). Bat echolocation Bouton, M. E. (2007). Learning and behavior: A contemporary synthesis. Sunderland, calls facilitate social communication. Proceedings of the Royal Society B: Bio- MA: Sinauer Associates. logical Sciences, 279(1748), 4827e4835. Bradbury, J., & Vehrencamp, S. (2011). Principles of animal communication. Sunder- Kulzer, E. (2005). Chiroptera. In Handbuch der zoologie [. In Handbook of zoology] land, MA: Sinauer Associates. (Vol. 3). Berlin, Germany: Walter de Gruyter. Braune, P., Schmidt, S., & Zimmermann, E. (2008). Acoustic divergence in the Lawrence, B. D., & Simmons, J. A. (1982). Measurements of atmospheric attenuation communication of cryptic species of nocturnal primates (Microcebus ssp.). BMC at ultrasound frequencies and the significance for echolocation by bats. Journal Biology, 6(1), 1e19. of the Acoustical Society of America, 71, 585e590. Brinkløv, S., Fenton, M. B., & Ratcliffe, J. M. (2013). Echolocation in oilbirds and Leippert, D. (1994). Social behaviour on the wing in the false vampire. Megaderma swiftlets. Frontiers in Physiology, 4,1e12. lyra. Ethology, 98(2), 111e127. Bukatko, D., & Daehler, M. W. (2003). Child development: A thematic approach (5th Liebermann, P., & Blumstein, S. E. (1991). Speech physiology, speech perception, and ed.). Boston, MA: Houghton Mifflin. acoustic phonetics. Camebridge, U.K.: Camebridge University Press. Catchpole, C. K., & Slater, P. J. B. (2008). Bird song: biological themes and variations. Li, Y., Wang, J., Metzner, W., Luo, B., Jiang, T., Yang, S., et al. (2013). Behavioral re- Cambridge, U.K.: Cambridge University Press. sponses to echolocation calls from sympatric heterospecific bats: implications Charlton, B. D., Huang, Y., & Swaisgood, R. R. (2009). Vocal discrimination of po- for interspecific competition. Behavioral Ecology and Sociobiology, 68(4), tential mates by female giant pandas (Ailuropoda melanoleuca). Biology Letters, 657e667. 5(5), 597e599. Maltby, A., Jones, K. E., & Jones, G. (2010). Chapter 2.4eUnderstanding the evolu- Cowen, R. (2005). History of life (4th ed.). Oxford, U.K.: Blackwell Publishing. tionary origin and diversification of bat echolocation calls. In M. B. Stefan (Ed.), Dechmann, D. K. N., Wikelski, M., Van Noordwijk, H. J., Voigt, C. C., & Voigt- Handbook of mammalian vocalizations (Vol. 19, pp. 37e47). Oxford, U.K.: Aca- Heucke, S. L. (2013). Metabolic costs of bat echolocation in a non-foraging demic Press. context support a role in communication. Frontiers in Physiology, 4,1e6. Mariappan, S., Bogdanowicz, W., Marimuthu, G., & Rajan, K. E. (2013). Distress calls Denzinger, A., & Schnitzler, H.-U. (2013). Bat guilds, a concept to classify the highly of the greater short-nosed fruit bat Cynopterus sphinx activate hypothalamic- diverse foraging and echolocation behaviors of microchiropteran bats. Frontiers pituitary-adrenal (HPA) axis in conspecifics. Journal of Comparative Physiology in Physiology, 4,1e15. A, 199(9), 775e783. Dorado Correa, A. M., Goerlitz, H. R., & Siemers, B. M. (2013). Interspecific acoustic Marshall, V. T., Schwartz, J. J., & Gerhardt, H. C. (2006). Effects of heterospecific call recognition in two European bat communities. Frontiers in Physiology, 4,1e8. overlap on the phonotactic behaviour of grey treefrogs. Animal Behaviour, 72(2), Duellman, W. E., & Pyles, R. A. (1983). Acoustic resource partitioning in anuran 449e459. communities. Copeia, 1983, 639e649. Maynard Smith, J., & Harper, D. (2003). Animal signals. Oxford, U.K.: Oxford Uni- Eimas, P. D., Siqueland, E. R., Jusczyk, P., & Vigorito, J. (1971). Speech perception in versity Press. infants. Science, 171(3968), 303e306. Mc Gregor, P. K., Falls, B. I., Catchpole, C. K., Pepperberg, I. M., Gerhardt, C. H., et al. Fantz, R. L. (1964). Visual experience in infants: decreased attention to familiar (1992). Playback and studies of animal communication. New York, NY: Springer. patterns relative to novel ones. Science, 146(3644), 668e670. Mendelson, T. C., & Shaw, K. L. (2012). The (mis)concept of species recognition. Fenton, M. B. (1985). Communication in the Chiroptera. Bloomington, IN: Indiana Trends in Ecology & Evolution, 27(8), 421e427. University Press. Monadjem, A., Taylor, P. J., Cotterill, W., & Schoeman, M. C. (2010). Bats of southern Fenton, M. B. (2003). Eavesdropping on the echolocation and social calls of bats. and central Africa: a biogeographic and taxonomic synthesis. Johannesburg, South Mammal Review, 33(3e4), 193e204. Africa: Wits University Press. Fitch, T. W. (2006). Production of vocalizations in mammals. In K. Brown (Ed.), Monteiro, A., & Podlaha, O. (2009). Wings, horns, and butterfly eyespots: how do Encyclopedia of language and linguistics (pp. 115e121). Oxford, U.K.: Elsevier. complex traits evolve? PLoS Biology, 7(2), e1000037. Grant, B. R., & Grant, P. R. (2010). Songs of Darwin's finches diverge when a new Neuweiler, G. (1990). Auditory adaptations for prey capture in echolocating bats. species enters the community. Proceedings of the National Academy of Sciences of Physiological Reviews, 70(3), 615e641. the United States of America, 107(47), 20156e20163. Neuweiler, G., Bruns, V., & Schuller, G. (1980). Ears adapted for the detections of Gregg, J. D., Dudzinski, K. M., & Smith, H. V. (2007). Do dolphins eavesdrop on the motion, or how echolocating bats have exploited the capacities of the echolocation signals of conspecifics? International Journal of Comparative Psy- mammalian auditory system. Journal of the Acoustical Society of America, 68(3), chology, 20(1), 65e88. 741e753. Guillen, J. (2012). FELASA guidelines and recommendations. Journal of the American Neuweiler, G., Metzner, W., Heilmann, U., Rübsamen, R., Eckrich, M., & Costa, H. H. Association for Laboratory Animal Science, 51(3), 311e321. (1987). Foraging behaviour and echolocation in the rufous horseshoe bat Guillen, A., Juste, J. B., & Ibanez,~ C. (2000). Variation in the frequency of the echo- (Rhinolophus rouxi) of Sri Lanka. Behavioral Ecology and Sociobiology, 20(1), location calls of Hipposideros ruber in the Gulf of Guinea: an exploration of the 53e67. adaptive meaning of the constant frequency value in rhinolophoid CF bats. Norberg, U. M., & Rayner, J. M. V. (1987). Ecological morphology and flight in bats Journal of Evolutionary Biology, 13,70e79. (Mammalia; Chiroptera): wing adaptations, flight performance, foraging 152 A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154

strategy and echolocation. Philosophical Transactions of the Royal Society of Schuller, G., & Suga, N. (1976). Storage of Doppler-shift information in the echolo- London B, 316(1179), 335e427. cation system of the CF-FM-bat, Rhinolophus ferrumequinum. Journal of Nosil, P., Crespi, B., & Sandoval, C. (2003). Reproductive isolation driven by the Comparative Physiology A, 105(1), 9e14. combined effects of ecological adaptation and reinforcement. Proceedings of the Seddon, N., & Tobias, J. A. (2010). Character displacement from the receiver's Royal Society of London. Series B: Biological Sciences, 270(1527), 1911e1918. perspective: species and mate recognition despite convergent signals in sub- Odendaal, L. J., & Jacobs, D. S. (2011). Morphological correlates of echolocation oscine birds. Proceedings of the Royal Society B: Biological Sciences, 277(1693), frequency in the endemic cape horseshoe bat, Rhinolophus capensis (Chiroptera: 2475e2483. Rhinolophidae). Journal of Comparative Physiology A, 197, 435e446. Seyfarth, R. M., Cheney, D. L., & Marler, P. (1980). Monkey responses to three Odendaal, L., Jacobs, D. S., & Bishop, J. (2014). Sensory trait variation in an echolo- different alarm calls: evidence of predator classification and semantic cating bat suggests roles for both selection and plasticity. BMC Evolutionary communication. Science, 210(4471), 801e803. Biology, 14(1), 60. Siemers, B. M., Beedholm, K., Dietz, C., Dietz, I., & Ivanova, T. (2005). Is species Page, R. A., Ryan, M. J., & Bernal, X. E. (2013). Be loved, be prey, be eaten. In identity, sex, age or individual quality conveyed by echolocation call frequency K. Yasukawa (Ed.), Case studies: Integration and application of animal behavior: in European horseshoe bats? Acta Chiropterologica, 7(2), 259e274. Vol. 3. Animal behavior (pp. 123e154). New York, NY: Praeger. Siemers, B. M., Kriner, E., Kaipf, I., Simon, M., & Greif, S. (2012). Bats eavesdrop on Pancratz, C. N., & Cohen, L. B. (1970). Recovery of habituation in infants. Journal of the sound of copulating flies. Current Biology, 22(14), R563eR564. Experimental Child Psychology, 9(2), 208e216. Sikes, R. S., & Gannon, W. L. (2011). Guidelines of the American Society of Mam- Pennetier, C., Warren, B., Dabire, K. R., Russell, I. J., & Gibson, G. (2010). ‘Singing on malogists for the use of wild mammals in research. Journal of Mammalogy, the wing’ as a mechanism for species recognition in the malarial mosquito 92(1), 235e253. Anopheles gambiae. Current Biology, 20(2), 131e136. Simmons, A. M., Popper, A. N., Fay, R. R., & Gerhardt, H. C. (2003). Acoustic Puechmaille, S. J., Borissov, I. M., Zsebok, S., Allegrini, B., Hizem, M., Kuenzel, S., et al. communication (Vol. 16). New York, NY: Springer. (2014). Female mate choice can drive the evolution of high frequency echolo- Slabbekoorn, H., & Smith, T. B. (2002). Bird song, ecology and speciation. Philo- cation in bats: a case study with Rhinolophus mehelyi. PLoS One, 9(7), e103452. sophical Transactions of the Royal Society of London Series B, 357(1420), 493e503. Rankin, C. H., Abrams, T., Barry, R. J., Bhatnagar, S., Clayton, D. F., Colombo, J., et al. South African National Standard. (2008). The care and use of animals for scientific (2009). Habituation revisited: an updated and revised description of the purposes, 10386 http://www.biologicalsciences.uct.ac.za/sites/default/files/ behavioral characteristics of habituation. Neurobiology of Learning and Memory, image_tool/images/75/files/SANS10386.pdf. 92(2), 135e138. Thabah, A., Rossiter, S. J., Kingston, T., Zhang, S., Parsons, S., Mya Mya, K., et al. Ruczynski, I., Kalko, E. K., & Siemers, B. M. (2007). The sensory basis of roost finding (2006). Genetic divergence and echolocation call frequency in cryptic species of in a forest bat, Nyctalus noctula. The Journal of Experimental Biology, 210(20), Hipposideros larvatus s.l. (Chiroptera: Hipposideridae) from the Indo-Malayan 3607e3615. region. Biological Journal of the Linnean Society, 88,119e129. Russ, J. M., Jones, G., Mackie, I. J., & Racey, P. A. (2004). Interspecific responses to Thomas, J. A., Moss, C. F., & Vater, M. (2004). Echolocation in bats and dolphins. distress calls in bats (Chiroptera: Vespertilionidae): a function for convergence Chicago, IL: University of Chicago Press. in call design? Animal Behaviour, 67(6), 1005e1014. Tinbergen, N. (1952). ‘Derived’ activities; their causation, biological significance, Russo, D., Mucedda, M., Bello, M., Biscardi, S., Pidinchedda, E., & Jones, G. (2007). origin, and emancipation during evolution. Quarterly Review of Biology, 27(1), Divergent echolocation call frequencies in insular rhinolophids (Chiroptera): a 1e32. case of character displacement? Journal of Biogeography, 34(12), 2129e2138. Tobias, J. A., & Seddon, N. (2009). Sexual selection and ecological generalism are Ryan, M. J., & Rand, A. S. (1993). Species recognition and sexual selection as a correlated in antbirds. Journal of Evolutionary Biology, 22(3), 623e636. unitary problem in animal communication. Evolution,647e657. Tobias, J. A., Planque, R., Cram, D. L., & Seddon, N. (2014). Species interactions and Sandoval, L., Mendez, C., & Mennill, D. J. (2013). Different vocal signals, but not prior the structure of complex communication networks. Proceedings of the National experience, influence heterospecific from conspecific discrimination. Animal Academy of Sciences of the United States of America, 111(3), 1020e1025. Behaviour, 85(5), 907e915. Vannoni, E., & Mcelligott, A. G. (2007). Individual acoustic variation in fallow deer Schnitzler, H.-U. (1968). Die Ultraschall-Ortungslaute der Hufeisen-Fledermause€ (Dama dama) common and harsh groans: a source-filter theory perspective. (Chiroptera-Rhinolophidae) in verschiedenen Orientierungssituationen. Zeits- Ethology, 113(3), 223e234. chrift für vergleichende Physiologie, 57(4), 376e408. Vignal, C., & Kelley, D. (2007). Significance of temporal and spectral acoustic cues for Schnitzler, H.-U., & Kalko, E. K. V. (2001). Echolocation by insect-eating bats. sexual recognition in Xenopus laevis. Proceedings of the Royal Society B: Biological BioScience, 51(7), 557e569. Sciences, 274(1609), 479e488. Schnitzler, H.-U., Moss, C. F., & Denzinger, A. (2003). From spatial orientation to food Vignal, C., Mathevon, N., & Mottin, S. (2008). Mate recognition by female zebra acquisition in echolocationg bats. Trends in Ecology and Evolution, 18(8), finch: analysis of individuality in male call and first investigations on female 386e394. decoding process. Behavioural Processes, 77(2), 191e198. Schuchmann, M., & Siemers, B. M. (2010). Behavioral Evidence for Community Wide Voigt-Heucke, S. L., Taborsky, M., & Dechmann, D. K. N. (2010). A dual function of Species Discrimination from Echolocation Calls in Bats. The American Naturalist, echolocation: bats use echolocation calls to identify familiar and unfamiliar 176,72e82. individuals. Animal Behaviour, 80(1), 59e67. Schuchmann, M., Puechmaille, S. B. J., & Siemers, B. R. M. (2012). Horseshoe bats West-Eberhard, M. J. (1983). Sexual selection, social competition, and speciation. recognise the sex of conspecifics from their echolocation calls. Acta Chiropter- Quarterly Review of Biology, 58(2), 155e183. ologica, 14(1), 161e166. Wilkins, M. R., Seddon, N., & Safran, R. J. (2013). Evolutionary divergence in acoustic Schuller, G., & Pollak, G. (1979). Disproportionate frequency representation in the signals: causes and consequences. Trends in Ecology & Evolution, 28(3), 156e166. inferior colliculus of doppler-compensating greater horseshoe bats: evidence for an acoustic fovea. Journal of Comparative Physiology A, 132(1), 47e54. Appendix

Table A1 Acoustic parameters (mean ± SD and minima and maxima in parentheses, N ¼ no. of individuals, n ¼ no. of calls) measured on calls in the test playback files

RcaDH85 (N¼9, n¼147) RcaHH84 (N¼6, n¼150) RcaTF86 (N¼10, n¼106) RcaLS75 (N¼6, n¼175) Rcl92 (N¼4, n¼223) Rda86 (N¼6, n¼105) 141 (2015) 101 Behaviour Animal / Jacobs S. D. Bastian, A.

Call duration (s) 0.043±0.008 (0.030e0.060) 0.041±0.008 (0.025e0.054) 0.047±0.007 (0.036e0.061) 0.036±0.010 (0.022e0.058) 0.047±0.019 (0.025e0.100) 0.041±0.007 (0.027e0.055) Interval (s) 0.111±0.048 (0.034e0.191) 0.103±0.071 (0.027e0.371) 0.137±0.042 (0.007e0.279) 0.086±0.056 (0.027e0.226) 0.067±0.030 (0.029e0.126) 0.131±0.055 (0.031e0.229) ± e ± e ± e ± e ± e ± e Distance to maximum 0.025 0.005 (0.016 0.036) 0.025 0.005 (0.014 0.033) 0.028 0.006 (0.014 0.039) 0.021 0.006 (0.009 0.039) 0.030 0.018 (0.014 0.082) 0.022 0.006 (0.012 0.033) frequency (s) Entropy 0.116±0.008 (0.107e0.137) 0.111±0.004 (0.107e0.120) 0.117±0.003 (0.112e0.124) 0.124±0.008 (0.112e0.14) 0.148±0.009 (0.134e0.165) 0.123±0.008 (0.113e0.146) Frequency fundamental 42.9±401 (42.5e46.8) 42.1±1050 (41.2e46.1) 42.6±259 (42.2e43.2) 37.3±726 (34.0e38.7) 45.8±329 (42.4e46.1) 42.4±439 (41.6e43.4) harmonic (kHz) Peak frequency second 84.7±442 (84.0e85.9) 83.9±442 (83.0e84.7) 85.3±513 (84.7e86.4) 75.4±625 (74.2e76.1) 91.9±438 (91.7e 92.0) 84.9±794 (83.7e86.6) harmonic (¼ RF) (kHz) Frequency third 124.2±2664 (112.5e125.0) 112.9±2156 (110.1e116.9) 121. 5±2006 (110.8e123.0) 113.2±927 (111.3e114.2) 137.7±835 (128.9e138.1) 116.1±861 (114.7e117.9) harmonic (kHz) Minimum frequency of 79.1±2882 (73.7e83.5) 81.4±1339 (78.3e83.0) 80.7±3283 (71.7e83.7) 69.8±2220 (65.6e74.4) 79.4±4499 (74.4e87.6) 76.1±3297 (71.5e82.7) initial FM component (second harmonic) (kHz) Minimum frequency of 74.1±4048 (66.60e81.8) 74.3±4092 (68.1e80.5) 71.5±1603 (68.6e75.9) 60.0±2945 (47.8e64.4) 65.3±2434 (60.3e70.5) 63.6±10687 (29.7e73.7) final FM component

(second harmonic) e (kHz) 154 Call rate (no./15 s) 9.8 10.0 7.1 11.7 14.9 7.0 153 154 A. Bastian, D. S. Jacobs / Animal Behaviour 101 (2015) 141e154

Table A2 List of behaviours shown by the bats during the habituation e dishabituation experiments and coded in the video recordings

Behaviour Description Category

Calm Bat shows no movement at all and hangs in a sleeping position on the perch Inactive/Calm Relocate/settle down Bat repositions its body usually accompanied by shaking slightly and ending in the sleeping position Active Scrambling Bat scrambles on the perch. Usually followed by settling down and ending in the sleeping position Active Off perch Bat leaves the perch, flying or scrambling Active Fly towards loudspeaker Bats flies off in a straight line towards the loudspeaker Attentive Grooming Bat grooms itself using its feet or licks its fur and wings Active Wing stretching partly Bat partly expands its wing(s) partly Active Wing stretching complete Bat completely expands its wing(s) Active Single ear tip twitch Bat slightly moves the tip of its ear(s) once Active Series of single ear tip twitches Bat slightly moves the tip of its ear(s) in a continuous series Active Ears twitching slowly Bat moves its entire ear(s) and the turns of the ear(s) are countable in real time Attentive Ear twitching rapidly Bat moves its entire ear(s). Movements are fast and single turns are not countable in real time Attentive Scanning Bat moves its entire ear(s) rapidly towards the sound source and has it head lifted up looking left and right Attentive Head raised slightly Bat lifts it head partly but the chin not visible Attentive Head raised completely Bat lifts its head completely looking up with its chin being visible Attentive Contraction of leg completely Bat contracts its leg(s) by bending its knees slightly Attentive Contraction of leg slightly Bat contracts it leg(s) completely by bending its knees fully Attentive